Weekly Robotics #382

Source custom parts with direct access to the manufacturers making them. No middleman, no black box. Jiga gives you a vetted network and one accountable partner standing behind every order. You know who's building your parts and communicate with them directly. AI-assisted operations handle supplier matching, quoting, quality verification, and production tracking, so the relationships stay human while the operations run at speed.

Excessive Overkill lives up to the channel name with this video. Their work creating an open-source FPGA controller for these arms is outstanding. This video is a crazy-good reverse-engineering story; you should watch it!

You need to answer the question in the title, but you have three minutes to answer it. What do you say? In this essay, Matthew T. Mason shows an interesting perspective.

I'll never get tired of the research coming from the ETH Robotic Systems Lab. This time, the team taught a robot to solve a gigantic Rubik's Cube. If you are wondering why, here are a few reasons I can think of: building SOTA perception and localization, fine-grained pedipulation (a new word I just learned), and interacting with the environment. To learn more about this work, check out this paper.

If you thought having quadruped robots solve Rubik's Cube was a work of art, I'm wondering what you would say about using a humanoid robot as a teleoperation platform to manipulate a surgical robot?

Joshua Bird has built a drone with just one motor. Using a hinge propeller hub and variable acceleration, the software can control the drone's pitch and roll. This feature is the first time I've heard of the PPA-CF filament, which seems to have some very interesting properties (and a price of 156 euros).

If you're having trouble keeping up with all the trends and framework du jour in the AI world, let me add one more to your plate: RoboJev is a model for labeling task success in robot learning. According to this thread, it's 48x cheaper and 28x faster than 6 Astra.

Citing the article:

Eka Robotics this week said it has raised $150 million in Series A funding. The startup said the new funding will help it scale its technology and customer deployments.

Founded in 2024, Eka is building a vision-force-action (VFA) model to develop an intuition of physics through reinforcement learning at scale. The Cambridge, Mass.-based company said that robots using its AI can pick a wide variety of objects outside the training distribution at a speed of roughly 1,500 operations per hour.